Fast Fingerprint Image Synthesis

This paper presents a fingerprint synthesis method that can generate a fingerprint with predefined minutiae points. Fingerprint type is chosen randomly and singular points positions and quantities are chosen randomly according to the fingerprint type. Orientation map is generated using fingerprint orientation model. Frequency map is generated. Initial image with drawn minutiae points that are oriented by orientation map is constructed. Iterative filtering of the initial image with Gabor filters that are oriented using orientation map and constructed using frequency map produces fingerprint image with minutiae points located at the predefined positions. An optimization of the iterative filtering is described. Synthetic fingerprint images are used to evaluate extraction algorithm's stability to noise. A measure of extraction algorithm’s robustness to noised fingerprint images is proposed.